Master's Theses



Degree Name

Master of Science (MS)


While the canonical nested subset pattern suggests that less species-rich areas will contain a proper subset of the species observed in richer areas, actual data sets do not show perfect nestedness; however, they show a pattern with more structure than would be expected by chance. Biological processes like immigration and extinction have traditionally been thought to produce the nested subset pattern. These processes acting indirectly could cause variation in the distribution and abundance of species that could produce nestedness at a variety of scales. Determining at what scales the pattern is observed might allow inference of processes that are more likely to be acting at those levels. My hypotheses were 1) species will be observed at more sites in the matrix when the matrix is constructed from sites in the center of the geographic range and fewer sites when the matrix is constructed from sites near the edge of the geographic range, 2) the matrix will be nested when sorted by incidence or abundance, 3) nestedness will be observed at intermediate scales, but not at the smallest or largest scales, and 4) the most species-rich site within the matrix will be the site of first occurrence for species. I used the 2003 North American Breeding Bird Survey data set (BBS) for the conterminous United States to construct the matrices used in analysis. An evenly spaced grid was used in a stratified sampling design to identify seed points. The geographic extent of analysis started within the route closest to the seed point and increased to include routes from multiple seed points. Two data matrices were constructed for each seed point at each scale; one sorted by incidence, and one sorted by total abundance of species as the independent variable. Due to limitations of existing software, I wrote a new program called Cudgel to test my hypotheses. Cudgel has the ability to sort the matrix by abundance or other independent variables. The metric N0, which calculates the number of unexpected absences, and null model Random 1, which preserves species presences while allowing presences to vary across sites, was used to calculate nestedness. The number of randomizations performed during the Monte Carlo simulation was 10,000. My results did not support the hypothesis that species will be observed at more sites in the matrix when the matrix is created from sites in the center of the geographic range. I did observe a significant correlation between incidence and abundance, with significant matrices showing a greater correlation (t = 2.976, df = 231, p = 0.003). Through graphical analysis, I determined that significance increased with increasing scale, but the metric values were lowest at intermediate spatial extents. While the species with the highest incidence were present at the most species-rich site first, species with low incidence showed greater variability in the site of first occurrence. Because there was a relationship between incidence and abundance, it is possible that the nested subset pattern could be caused by distribution and abundance. If the nested subset pattern is caused by distribution and abundance, research should be focused on the factors that are producing the observed variation in distribution and abundance.


Dr. Rob Channell

Date of Award

Spring 2010

Document Type



© 2010 Elita Baldridge


For questions contact

Included in

Biology Commons